Incorporating Uncertainty within Human-in-the-Loop Analytics for Data Exploration

Friday, April 6, 2018 - 3:30pm

Speaker(s): 
Leanna House, Virginia Tech

Abstract: 

While we are inundated with big data, we are also inundated with uncertainty. There is uncertainty in data collected, uncertainty in the science motivating the data collection, uncertainty in methods used to summarize data, and uncertainty in judgements formed from data summaries.  Alas, when exploring data, analysts often avoid quantifying formally and/or communicating degrees of uncertainty in data summaries.  There are several potential reasons for avoiding uncertainty, ranging from difficulty (it is hard to model big data well) to misplaced confidence in laws of large numbers to fear of inconsistencies in how humans process uncertainty.  In this talk, we take a visual analytic and probabilistic approach to engage humans in learning from big data visually, while considering uncertainty. Specifically, we start with a method we developed called Bayesian Visual Analytics (BaVA) and incorporate novel, visual metaphors of uncertainty, in the context of weighted multi-dimensional scaling visualizations. This presentation may open doors to future research into how humans process uncertainty as they visually explore and learn from big data.

Seminars generally take place in 116 Old Chemistry Building on Fridays from 3:30 - 4:30 pm. For additional information contact: karen.whitesell@duke.edu or phone 919-684-8029. Sorry, but we do not have reprints available. Please feel free to contact the authors by email for follow-up information, articles, etc. Reception following seminar in 211 Old Chemistry

Old Chemistry 116

Location Info